Lachlan,
As data may be poor for many reasons, results may be unstable in
many ways. In my experience, elimination of the restraints in the
final refinement cycles gives always some change in parameters,
most time small but sometimes it happens that a homeless atom
begins to wander/vanish etc, all without an improvement/change of
R. In a few cases, I observed significant changes (both decrease
and increase) of R when releasing the restraints. I never published
such stuff, probably there's something wrong with the model or the
data.
Anyway, my use for restraints is mainly as a strategy during
refinement, e.g. for group-subgroup transitions, and during the last
stage I try to release them all together, except for symmetry
restraints which should be kept as should an inversion center, for
example. Just for the meaning of the esds!
best regards
Miguel
On 15 Apr 01, at 4:21, Lachlan Cranswick wrote:
>
> For people using bond-restraints on medium to large inorganic/zeolite
> structures to keep things physically reasonable during Rietveld
> refinement (for data of moderate quality): are there any comments on
> how R-factors tend to vary if restraints are released - or set to near
> zero values.
>
> Are there negligible drops in R-factors as atoms start wandering away
> on a holiday in unrestrained refinements vs restrained refinements?
> (this assumes the model is basically correct)
Miguel Gregorkiewitz
Dip Scienze Terra, Università
via Laterina 8, I-53100 Siena, Europe
fon +39'0577'233810 fax 233938
e-mail [EMAIL PROTECTED]
